Open BTS, a GSM experiment in rural Zambia Jacqueline Mpala1 and Gertjan van Stam2 1

2

University of Zambia, Lusaka, Zambia [email protected]

Nelson Mandela Metropolitan University, Port Elizabeth, South Africa [email protected]

Abstract. This paper presents an OpenBTS implementation case study in rural Zambia. The work focused on applied research and development of a cost effective telecommunication system to provide mobile communications in a rural village. The system aimed to test appropriate technologies involving limited hardware needs, incurring low capital expenditure, and with limit operational costs. The practical experience revealed that nontechnical aspects like logistics of equipment availability, internet bandwidth constraints, and interaction with the frequency regulator are the main implementation constraints. Key words: openBTS, Zambia, low-cost communication, rural area networks, mobile telephony

1 Introduction Access to information is a basic human right. Article 20(1) of the Zambian constitution states that a person shall not be hindered in the enjoyment of his freedom of expression which includes freedom to receive ideas and information [1]. In Zambia, the market for Information and Communication Technologies (ICT) and services grows significantly. A survey depicts mobile communication growing from 464,000 in 2004 to 4.4 million users in 2009 [2]. The liberalization of the ICT sector in Zambia facilitates this increase in usage of ICT [3]. Growth in ICT access facilitates economic growth in Zambia [4]. Liberalization of the telecommunications market results in increased competition. This contributes to lower access and user costs [2]. The International Telecommunications Union (ITU) mentions Zambia among the top ten countries with decreasing costs for ICT. However, in Zambia, real costs of ICT are still among the highest in the world [5]. Although there are large differences in terms of urbanization in Africa, on average in developing countries the vast majority of people (approximately 70% to 85% of the labor force) live in rural areas. 93.9% of Zambia’s surface is customary, rural land [6] and 61.2% of Zambia’s population lives in rural areas [7].

2

Mpala et al.

It is essential for rural areas to develop, so new opportunities emerge and innovations occur [8]. Opportunities and innovations provide for increase in productivity and for the populace to enjoy an improved quality of life. Chief Chikanta and Mweetwa mention that rural areas face significant resource challenges in the form of poor network coverage, limited transport, electricity, and water supply. They deem a disparity in access undesirable, as it demonstrates that people of different communities have unequal opportunities to benefit from technology in their daily lives. The ICT sector in Zambia lacks sufficient people with adequate ICT skills. Zambia needs people with such skills to drive the nations envisioned progress towards a knowledge economy by 2030 [9][4]. Habeenzu cites a survey on ICT industry skills that reports the figure of 300 people with graduate qualifications in ICTs in Zambia in 2008. He contributes the low number of skilled persons to the lack of adequate formal ICT training facilities at tertiary education level in Zambia. Low numbers of engineers are a common occurrence in developing countries [10]. Though there has been a considerable reduction in the cost of access to mobile phones, other factors such as geographical challenges, lack of supporting infrastructures, and limited abilities to recognize opportunities play a role in the availability of mobile services in rural areas. The World Bank observes that mobile network operators find it commercially infeasible to operate in rural areas [11]. Zambian rural areas are usually not connected to the electricity grid. They encapsulate sparsely populated areas and poor road infrastructure. Rural areas have always lagged in development, including limited availability of ICT. Deploying traditional GSM networks in rural areas is inefficient as villages are much dispersed, typically separated by kilometers of uninhabited areas. In developing nations, more people own mobile phones than personal computers. In 2011, the National Geographic magazine reports that the world’s 1 billion poorest own 22 mobile phones per 100 people, but only have 1.2 computers per 100 people [12]. The yearly income of people at ’the bottom of the pyramid’ is USD 1,000 or less, and most reside in Africa. Lower-middle income populations, which account for 4 billion people, earning between USD 1,000 and 4,000 per year, living mostly in India, China, Southeast Asia, and parts of Africa, own 47 mobile phones per 100 people and have 4.3 personal computers per 100 people. Coming from this setting, this paper about an OpenBTS study in rural Zambia, presents an overview of the specific, Zambian environment (section 2), related work (section 3), and a description of the works and the findings (section 4 and 5). A discussion of this first practical implementation experiment fills section 6, and the conclusions feature in section 7.

Open BTS, Zambia

3

2 Environment The OpenBTS case study is part of the collaboration per Memorandum of Understanding, by the University of Zambia and LinkNet. LinkNet is a rural based, co-operative organization that provides internet connectivity in rural areas of Zambia [13]. LinkNet is an operational unit of Macha Works. Macha Works’ vision is to inspire people in rural communities to reach their collective and individual potential. The co-operation of UNZA and LinkNet facilitate enhancing educational and learning facilities with a focus on rural areas. It aligns with national key objectives in human capacity building, research and development, and innovation in Zambia’s ICT sector [14]. LinkNet’s network operations center is in the rural community of Macha, in the Southern Province of Zambia. Macha Works facilitates community driven projects, including internet connectivity, in seven villages in Zambia. These sites are Chilonga (Northern Province), Chitokoloki, Kalene, and Mukinge (North Western Province), Minga (Eastern Province), and Chikanta and Macha (Southern Province). In each site, LinkNet packages ICT functionalities using a Linknet Resource Container (LRC) base. A LRC is a locally-refurbished shipping container. It hosts a VSAT outdoor unit for satellite connection to the Internet, an uninterrupted power supply (UPS), a network server and gateway, user-PCs, and wireless area network connections. It provides a secure area with PCs for training and community purposes. Solar panels supply power to two LRCs (Chikanta and Chitokoloki), while the other LRCs connect with the national electricity grid. In 2011, mobile phone, GSM, coverage is adequate at 4 out of 7 LinkNet locations in Zambia. Two locations have limited GSM coverage, where one has to search for ’catching a signal’ from a distant transmitting point. Up to the beginning of 2012, Chikanta, a village located 100 kilometer northeast of Kalomo did not have GSM coverage. The nearest GSM tower stands at a distance of 40 kilometers. Macha Hospital’s catchment area involves operations of 12 rural health centers. In turn, each rural health center caters for a number of rural health posts in its own area. Not all these health centers have mobile coverage, and fewer of the health posts are within the range of a mobile signal. Even with patchy, limited coverage, the use of GSM is advantageous for health interventions. For instance, Macha Research Trust reports on the use of SMS to send weekly information of rapid malaria tests used, and number of positive diagnoses to Macha [15].

3 Related Work The OpenBTS is a UNIX/Linux based, free and open source software (FOSS) that configures Software Defined Radios as an independent GSM access point. The system allows GSM handsets to connect, and facilitate call set-up. OpenBTS integrates the GSM-network functionalities of traditional, high cost Base Tranceiver Station (BTS), Base Switching Center (BSC) and Mobile Switching Center (MSC) into one low cost platform. Calls route through an Internet Protocol (IP)

4

Mpala et al.

data network with an Asterisk Private Branch Exchange (PBX), using Voiceover-IP (VoIP) through Session Initiation Protocol (SIP), instead of call traffic through a GSM switch (the MSC). OpenBTS necessitates software radio hardware known as Universal Software Radio Peripheral (USRP). The cost of a USRP is about USD 700. With generic GSM to SIP conversion behind the air interface, OpenBTS interfaces with standard IP components. The USRP presents a GSM air interface (Um) to standard GSM handset and uses the Asterisk software PBX to connect calls [16]. The combination of the global-standard GSM air interface with low-cost VoIP backhaul forms the basis of the cellular network, with deployment and operational costs at substantially lower level than existing technologies in many applications, including rural cellular deployments and private cellular networks in remote areas [16], [17]. During 2011, Anand et al. did OpenBTS experiments in a USA based laboratory environment. These experiments focused on preparation of further OpenBTS research for rural deployment, in collaboration with LinkNet/UNZA [18]. In their paper, Anand et al expand on the rationale for OpenBTS, and describe the technical performance that can be expected in a mixed traffic environment, using traffic patterns observed in Macha. Other work mentions the Village Base Station as a generic concept [19], or providing input for financial business planning [20]. Kretchmer et al. evaluated the Quality of Service of OpenBTS mobile calls across a multi-hop wireless testbed that carries typical Internet traffic [21].

4 Study Design The case study focused on an actual implementation of OpenBTS in rural Africa. The aim of the study was to deploy and review such implementation by monitoring its deployment and usability by establishing real communication of mobile phones with each other and the outside world. The study aimed to provide a basis for replication of OpenBTS tests, by running experiments in the rural area of Chikanta, in Zambia’s Southern Province. The implementation involves a small base transceiver station utilizing OpenBTS software. This Linux-based software application configures the universal software radio peripheral (USRP) to present a GSM air interface to standard GSM phones and interfaces with the Asterisk Voice over Internet Protocol Private Branch exchange (VoIP PBX) to connect calls. OpenBTS implementations can connect with each other and to the internet via standard IP technologies and backhaul, like Wifi or Wimax. Such implementation differs substantially from conventional GSM BTS implementations. A standard BTS routes calls via a MSC and its dedicated back-bone network. The project did not address billing system integration. The research focused on a deployment, with prime investigation done in a rural Zambian community by Zambian researchers. This is the first exercise in rural Zambia as far as we know.

Open BTS, Zambia

5

For practical support, the research linked up with computer science students in Germany. They work on a project to combine a low cost GSM base station with a Wi-Fi mesh network. Relationships also exist with computer science students in the USA. Practical deployment of the implementation took place between October 2010 and July 2011. It involved practical implementation of a low power, OpenBTS implementation, aimed at establishing and testing calls using mobile phones in Zambia. Notes on the project execution and experiences provide information on challenges posed by many forms of constraints.

5 Findings The activities succeeded in establishing calls between and from mobile telephones in Zambia. However, to achieve this, a whole range of physiological, legal, and practical challenges had to be overcome. Equipment required by OpenBTS is not for sale in Zambia. International purchase of equipment was challenging since most suppliers requested payment by credit cards. Credit cards are not available in Zambia. At the time of the research, VISA debit cards were entering the urban Zambian market for the first time. However, none of the research team members possessed a debit card or an (international) credit card. International collaborators intervened. They purchased the equipment abroad and facilitated processing of the payment to the supplier. Final payment involved a cash transfer from the project to the collaborator. Logistics in transport and importing of equipment raised significant hurdles. After much trial and error, a LinkNet research partner in Germany stepped in. They sourced the test equipment and brought it in person from Germany to a conference in South Africa. From there, the equipment was flown to Zambia. Rural villages are far apart, and the roads are dangerous. The project team travelled frequently between urban and rural areas for discussions with the community and regulator and to deliver progress reports to various institutes involved, and for actual testing. At one of these journeys, while returning from project demonstrations and community-capacity building activities in Chikanta, the author and IEEE colleagues had a road traffic accident, after hitting a bump on a gravel road. Several people needed medical attention in the rural hospital, and one car was destroyed. The research partner in Germany preconfigured the equipment. Upon arrival, it worked for tests establishing local calls. The test system consisted of a low power PC and an USRP. On site, the addition of an additional harddrive added more capacity to the system. This increase in capacity facilitated the expansion of functionalities. Configuring the increased capacity of the equipment took much unexpected effort. Core software and various driver upgrades for the USRP and Asterisk had to be downloaded from project sites on the internet. This requires broadband internet connectivity. Although the rural project site in Macha is relatively well connected by Zambian standards (Committed Information Rate

6

Mpala et al.

256 kbs, bursting up to 1028 kbs, discounted cost USD 1,400 per month), uninterrupted facilitation of the large file downloads provided became an immense challenge. The sheer number of files that needed to be downloaded, issues of shared bandwidth use, frequent electricity outings, and international satellite bandwidth interruptions contributed to the challenges. Downloads took place during the nights and at weekends, trying to limit the effect of the download on operational use of the network in Macha. It took several weeks to complete all downloads. After concluding the initial configurations, the test frequencies were set. The Zambia Information and Communications Technology Authority (ZICTA), regulates the utilization of radio transmissions. This institute was established through the Information and Communications Act in 2009 [22]. Its mandate includes service and supplier’s licenses, provision of radio services, settings and allocation of frequency spectrum, numbers and electronic addresses, and interconnection. ZICTA is also the implementer of universal service programs. Gaining permission from ZICTA took many meetings and much travel from the rural area (test site in Macha) to the urban area (ZICTA offices in Lusaka). As all concepts involved in the project were new to Zambia, contents of the meetings involved detailed sensitization and explanation of underlying engineering concepts, explanation and providing proof of the research and development nature of the activities, and brainstorming sessions as how to facilitate and proceed. OpenBTS operates in the GSM frequency bands. Zambia does not have a legal framework for test frequencies for research in ICT. Unlike, US’ regulator FCC granting experimental licences for radiated power levels lower than 8 Watt, there is no regulation regarding the allotment of test frequencies for pilots in Zambia. The commercial nature of GSM frequencies rendered ZICTA apprehensive of allocating frequencies for the OpenBTS tests. After many meetings, ZICTA allocated the project temporarily frequencies in the extended GSM-900 range. To assure close supervision of the use of tests, ZICTA requested activities to take place under their scrutiny on site at ZICTA’s premises in Zambia’s capital city Lusaka. As a consequence, equipment travelled thousands of kilometers within Zambia. Internet access at ZICTA’s offices is limited. Actual tests were difficult to perform due to limited access and bandwidth there. Within the OpenBTS coverage area, calls can be made even when the internet link is not available. For calls to other networks there is the necessity for a (relatively) solid internet connection, which was available in Macha, but not at ZICTA offices in Lusaka. The quality of calls were highly dependent on the speed of the internet connection. The OpenBTS set-up in rural Macha varied, depending on congestion and latency of the VSAT connection. Especially in the highly congested lines at ZICTA, delays became great and hampered conversations. Tests showed that a number of no-brand handsets sold ’on the streets’ in Zambia are incompatible with the system.

Open BTS, Zambia

7

Calls from within the OpenBTS can be routed via VoIP service providers. When using a VoIP gateway, payment for such services necessitates international payments mostly involving credit cards, or otherwise by international transfer, both of which proved a hurdle for use. As mentioned, credit cards do not exist in Zambia. Facility Management centers regularly block network or service access for African IP addresses. Also, use of debit or credit card from African IP addresses regularly fail, due to settings that restrict services provided to African IP numbers. Depending on the availability of suitable international internet connectivity, it proved feasible to establish high quality connections with western based VoIP or Public Switched Telephony Network (PSTN) users. However, routing calls from Zambia to Zambia, using existing VoIP providers, are expensive and offer low quality. No VoIP-to-PSTN gateway service provider exists in Zambia.

6 Discussion The project was the first practical implementation experiment with OpenBTS in Zambia. Practicalities involving sourcing of equipment, downloading of software, and gaining permission to use radio frequencies, took months to complete. These practical challenges were intertwined with many logistical issues and travel hazards. These non-technical project aspects took most of the research efforts (cf [23]). A pre-requisite for the acquisition of test frequencies was a close relationship with the regulator. Further, there is a need for healthy relationships with the community for the implementation of test deployments. Constant communication with the wider community in a culturally acceptable manner is another prerequisite. Building such relationships took much effort and financial resources, and many days of travelling. Although the project was able to establish GSM calls, the quality of the internet backbone proved to be of prime importance for a feasible deployment of OpenBTS. Trunk calls use routings over the IP/internet network. Adding more traffic on an already congested, limited bandwidth internet link in rural areas degrades the performance of the system. When international internet capacity was sufficient, the establishment of connections with western receipients proved achievable. However, reaching a Zambian telephone recipient proved difficult, involving challenging financial interactions, high costs and low quality connections. Use of OpenBTS showed to have considerable limitations: 1. the system does not support data transmission, i.e. does not support GPRS; 2. the system does not necessarily operate with all phones that sell in the Zambian market; 3. OpenBTS did not support roaming, and there is no handover of calls between base stations (an issue currently being addressed).

8

Mpala et al.

7 Conclusion and Future Work OpenBTS is a viable and low cost system to build a local rural cellular GSM network. It can be used in sites when enough internet bandwidth is available. Local calls can be made even if an internet link is not available. Though OpenBTS comes with clear limitations, it shows promising potential for rural mobile telecommunication. It provides a solution where low-cost, basic communication functions suffice. The case study provides insight into nonengineering aspects that influence implementing an OpenBTS testbed in rural Zambia. Universal Service Funds aim at encouraging service providers to invest in rural areas. OpenBTS is an innovation that could be implemented in rural areas as a stand-alone system. Under the banner VillageCell [18], and based upon the relationships established, and experiences gained by this project, future work envisions deriving a simple and easy system architecture for localized cell phone communication software involving defined radios (SDRs) and open-source solutions (OpenBTS and Asterisk) to further explore the opportunities for low-cost alternatives to high-end cell phone networks. Propsals for implementation of OpenBTS for rural communications are being submitted for universal access funding by ZICTA.

8 Acknowledgement Authors acknowledge the input and feedback of Mr. A.C. Nalubamba (Zambia), LinkNet staff at Macha for facilitation and Linux training, and Peter Hasse (Fraunhofer FOKUS, Germany) for much practical help. Further gratitude to Tim Patton (UK) for unpacking OpenBTS, Hanna Weijers (NL), Karl Jonas (Germany), and David Johnson (USA) for scientific support and logistics, and Mr. M. Phiri, IEEE Zambia, for all assistance rendered. We thank the reviewer for valuable comments and suggestions. IEEE’s Engineering Projects in Community Service (EPICS) provided seed funding for this project.

References 1. Government of the Republic of Zambia. The Constitution of the Republic of Zambia (1996) as amended by Act No. 18 of 1996. 1996. 2. David Montez. Mobile Communications in Zambia A demand-side analysis based on the AudienceScapes Survey, including insights on mobile money use. Number October. InterMedia Survey Institute, Washingtion, DC, 2010. 3. Ministry of Commerce Trade and Industry. Investment Policy Review of Zambia. Advancing investment policy reform. Government of the Republic of Zambia, Lusaka, 2011. 4. Government of the Republic of Zambia. ICT Policy Zambia. 2005.

Open BTS, Zambia

9

5. ITU-D. Measuring the Information Society. ITU, Geneva, 2011. 6. Martin Adams. Land tenure policy and practice in Zambia: issues relating to the development of the agricultural sector. Mokoro Ltd, Oxford, 2003. 7. Central Statistics Office Zambia. 2010 Census of Population and Housing Preliminary Report, 2011. 8. His Royal Highness Chief Chikanta and Fred Mweetwa. The Need for Information and Communications Technologies. LinkNet, 2007. 9. Shuller Habeenzu. Zambia ICT Sector Performance Review 2009/2010, volume Two. Research ICT Africa, 2010. 10. UNESCO. Engineering: Issues Challenges and Opportunities for Development. UNESCO Publishing, Paris, 2010. 11. World Bank. Information and Communications for Development 2012: Maximizing Mobile. World Bank, Washington, DC, 2012. 12. Jill Duffy. 7 Billion People: Who Owns the Computers and Cell Phones? PC Magazine, 2011. 13. Karel Matthee, Gregory Mweemba, Adrian Pais, Gertjan van Stam, and Marijn Rijken. Bringing Internet connectivity to rural Zambia using a collaborative approach. In ICTD 2007. Ieee, 2007. 14. Government of the Republic of Zambia. National Policy on Science and Technology. 1996. 15. Aniset Kamanga, Petros Moono, Gillian Stresman, Sungano Mharakurwa, and Clive Shiff. Rural health centres, communities and malaria case detection in Zambia using mobile telephones: a means to detect potential reservoirs of infection in unstable transmission conditions. Malaria Journal, 9(96), 2010. 16. David A Burgess and Harvind S Samra. The Open BTS Project. 2008. 17. OpenBTS. The OpenBTS Project. 18. Abhinav Anand, David L Johnson, and Elizabeth M Belding. VillageCell: Cost Effective Cellular Connectivity in Rural Areas Categories and Subject Descriptors. In ICTD 2012, 2012. 19. Kurtis Heimerl and Eric Brewer. The Village Base Station. Proceeding NSDR ’10 Proceedings of the 4th ACM Workshop on Networked Systems for Developing Regions Systems, pages 5–6, 2010. 20. Carlos Rey-moreno, Zukile Roro, Masbulele Jay Siya, Javier Simo-reigadas, Nicola J Bidwell, and William D Tucker. Towards a Sustainable Business Model for Rural Telephony. In III International Conference on Research in ICT for Human Development (JITIC4DH), Cuzco, 2012. 21. Mathias Kretschmer, Peter Hasse, Christian Niephaus, Thorsten Horstmann, and Karl Jonas. Connecting Mobile Phones via Carrier-Grade Meshed Wireless Back-Haul Networks. E-Infrastructures and E-Services on Developing Countries. Africomm 2010, 2010. 22. Government of the Republic of Zambia. The Information and Communication Technologies, Act No. 15 of 2009, 2009. 23. Eric Brewer, Michel Demmer, Melissa Ho, R.J. Honicky, J Pal, M. Plauche, and Sonesh Surana. The Challenges of Technology Research for Developing Regions. IEEE Pervasive Computing, 5(2):15–23, April 2006.